A support vector machine approach to identification of proteins relevant to learning in a mouse model of Down Syndrome

Abstract

Down Syndrome is a fairly common disorder causing intellectual disability. Although there is no treatment for the learning difficulties associated with Down Syndrome, the drug memantine has been shown to improve learning ability in a Down Syndrome model of mice (Ts65Dn) exposed to Context Fear Conditioning (CFC), an existing technique used in determining the learning ability of normal mice which does not normally produce results in Ts65Dn mice. It remains to be seen, however, precisely how memantine increases the learning ability of Ts65Dn mice exposed to CFC. This work seeks to contribute to this knowledge by analyzing the expression of 77 proteins obtained from normal and Ts65Dn mice, with and without memantine and with and without exposure to CFC. A machine learning method known as a Support Vector Machine, which works by building classifiers to distinguish between the various sets of mice based on protein levels, is used for the analysis. Feature selection is then used to choose the proteins whose levels appear to play a significant role in the classification. For all classifiers examined, this method appears to outperform methods that have previously been used for analyzing the data. The results of the classifiers have the potential to be utilized in further biological study for detection of protein response patterns relevant to learning in normal mice and in Ts65Dn mice treated with memantine.